August 06, 2004

Poll Dancing

I hate to drag out my favorite whipping boy again, but I just saw Michael Moore on Leno, talking about a Harris Online poll. The gist of it was (as was printed in a lot of articles right after the poll came out) that those darned republicans better watch out, because the poll said...well, I'll let the Harris article tell you:

Also bad news for the president is that significant minorities of Republican viewers of the movie give it positive ratings (44%) and think it treats the president fairly (30%).

Well, you know me and polls. As the grizzled old man down the road from me always used to say "Polls is the devil's tools...but poles is the barge pilot's tools." He was always full of nuggets of wisdom like: "Tornado got Ol' Yeller, git to the cellar," and "git outta ma yard, or I'll brain you with a mallet." But I digress.

Those percentages didn't sound right to me, so I decided to take a little look at the numbers for myself.

The first thing I noticed is that everything is percentages; I couldn't find the actual numbers anywhere. That made me a little suspicious. Anybody out there with crazy math skills, feel free to see if you can figure them out for me. I'd also be interested if anyone can find them online somewhere. I checked the Harris site with no luck. But I'll work with the percentages, and some (semi) educated guesswork.

Total number of participants: 2,242 adults. A good number, I guess. Now let's suppose that a third of those people identified themselves as Republican--about 747 people. That percentage may seem a little low, but I'll go through it again in a second with what I feel is an upper limit, so we can get a range.

According to the poll, 7% of those Republicans actually saw the film (I won't even mention the +-2 percentage point MOE). That cuts it to 52 people. Of those 52 Republicans who saw the film, it is true that 44% gave it a positive review. That's 23% excellent--12 people, and 21% good--11 people. 30% of them thought the film was fair to President Bush. That's 16 people. So this significant minority they're talking about? 16 people out of 2242? That's about 7/10 of a percent. I don't consider that "significant."

But maybe assuming that a third of the people are Republicans is a little low. Okay, I'll run through it again, but this time I'll assume that fully half of respondents identified themselves as republicans--1,121. I'm sure you'll all agree that number is much higher than likely. Now...
Republicans who saw the film? 78.
who thought it was excellent? 18.
who thought it was good? 16
who thought it was fair to President Bush? 23. (about 1%)

Either way, we're looking at less than one percent of those who responded to the poll. I wonder if that's keeping Bush up nights...

p.s. All of this depends upon my math skills, which used to be somewhat sharp, but now are used only to figure out my odds of drawing to an inside straight. I think I've done okay here, but please feel free to check the numbers.

UPDATE: The post above cites another survey that seems to reinforce what I've written here. In that survey, the makeup of those who had seen the film were as follows:
55% - Democrats; 24% - Independents (although based on other questions, the study labeled them as "much more liberal than independents at large."); and 10% Republicans. I don't know about the other 11%.

I was interested to see how substituting a similar breakdown to the Harris Poll would affect the numbers:
13% of 2242 (those who had seen the film) - 291
10% of 291 (Republicans who had seen the film) - 29
23% of 29 (Republicans who thought it was excellent) - 7
21% of 29 (Republicans who thought it was good) - 6
30% of 29 (Republicans who thought it treated Bush fairly) - 9

Granted, those may not be the actual numbers, but I feel pretty comfortable going with the other study's breakdown. They also used a sample over twice as large as the Harris poll, but their percentage of respondents who had seen the film was a little lower at about 8.5%.

No comments:

Post a Comment